analyses are consistent with the concept of expected utility maximization. Given that no model is true, the … negative variance! Thus we should replace the linear probability model by another model which makes sense! … coefficient @ in the standard linear model is time-invariant. Thus, the linear regression is described by one